Dataset Page: apply markdown transforming only to TEXTBOX type fields

package.json

@@ -49,6 +49,7 @@
     "react-toastify": "11.0.2",
     "react-topbar-progress-indicator": "4.1.1",
     "sass": "1.58.1",
+    "turndown": "7.2.0",
     "typescript": "5.7.2",
     "use-deep-compare": "1.2.1",
     "vite-plugin-istanbul": "4.0.1",

src/dataset/domain/repositories/DatasetRepository.ts

@@ -6,7 +6,12 @@ import { DatasetsWithCount } from '../models/DatasetsWithCount'
 import { VersionUpdateType } from '../models/VersionUpdateType'
 
 export interface DatasetRepository {
-  getByPersistentId: (persistentId: string, version?: string) => Promise<Dataset | undefined>
+  getByPersistentId: (
+    persistentId: string,
+    version?: string,
+    requestedVersion?: string,
+    keepRawFields?: boolean
+  ) => Promise<Dataset | undefined>
   getLocks(persistentId: string): Promise<DatasetLock[]>
   getByPrivateUrlToken: (privateUrlToken: string) => Promise<Dataset | undefined>
   getVersionDiff: (

src/dataset/domain/useCases/getDatasetByPersistentId.ts

@@ -4,9 +4,13 @@ import { Dataset } from '../models/Dataset'
 export async function getDatasetByPersistentId(
   datasetRepository: DatasetRepository,
   persistentId: string,
-  version?: string
+  version?: string,
+  requestedVersion?: string,
+  keepRawFields?: boolean
 ): Promise<Dataset | undefined> {
-  return datasetRepository.getByPersistentId(persistentId, version).catch((error: Error) => {
-    throw new Error(error.message)
-  })
+  return datasetRepository
+    .getByPersistentId(persistentId, version, requestedVersion, keepRawFields)
+    .catch((error: Error) => {
+      throw new Error(error.message)
+    })
 }

src/dataset/infrastructure/repositories/DatasetJSDataverseRepository.ts

@@ -172,10 +172,11 @@ export class DatasetJSDataverseRepository implements DatasetRepository {
   getByPersistentId(
     persistentId: string,
     version: string = DatasetNonNumericVersion.LATEST_PUBLISHED,
-    requestedVersion?: string
+    requestedVersion?: string,
+    keepRawFields?: boolean
   ): Promise<Dataset | undefined> {
     return getDataset
-      .execute(persistentId, version, includeDeaccessioned)
+      .execute(persistentId, version, includeDeaccessioned, keepRawFields)
       .then((jsDataset) => this.fetchDatasetDetails(jsDataset, version))
       .then((datasetDetails) => {
         return this.fetchDownloadSizes(persistentId, version).then((downloadSizes) => {
@@ -223,7 +224,8 @@ export class DatasetJSDataverseRepository implements DatasetRepository {
         return this.getByPersistentId(
           persistentId,
           DatasetNonNumericVersion.LATEST_PUBLISHED,
-          (requestedVersion = version)
+          (requestedVersion = version),
+          keepRawFields
         )
       })
   }

src/metadata-block-info/domain/models/MetadataBlockInfo.ts

@@ -101,7 +101,7 @@ export interface MetadataBlockInfoDisplayFormat {
 
 export type MetadataBlockInfoDisplayFormatFields = Record<string, MetadataFieldInfo>
 
-export type MetadataFieldInfo = Pick<MetadataField, 'displayFormat'>
+export type MetadataFieldInfo = Pick<MetadataField, 'displayFormat' | 'type'>
 
 export const METADATA_FIELD_DISPLAY_FORMAT_PLACEHOLDER = '#VALUE'
 export const METADATA_FIELD_DISPLAY_FORMAT_NAME_PLACEHOLDER = '#NAME'

src/metadata-block-info/infrastructure/mappers/JSMetadataBlockInfoMapper.ts

@@ -23,7 +23,7 @@ export class JSMetadataBlockInfoMapper {
     const fields: MetadataBlockInfoDisplayFormatFields = {}
 
     Object.entries(jsMetadataBlockInfoFields).forEach(([key, value]) => {
-      fields[key] = { displayFormat: this.toDisplayFormat(value.displayFormat) }
+      fields[key] = { displayFormat: this.toDisplayFormat(value.displayFormat), type: value.type }
 
       if (value.typeClass === 'compound' && value.childMetadataFields) {
         this.processCompoundFields(value.childMetadataFields, fields)
@@ -38,7 +38,7 @@ export class JSMetadataBlockInfoMapper {
     result: MetadataBlockInfoDisplayFormatFields
   ): void {
     Object.entries(jsFields).forEach(([key, value]) => {
-      result[key] = { displayFormat: this.toDisplayFormat(value.displayFormat) }
+      result[key] = { displayFormat: this.toDisplayFormat(value.displayFormat), type: value.type }
     })
   }
 

src/sections/dataset/DatasetProvider.tsx

@@ -26,7 +26,13 @@ export function DatasetProvider({
 
   const getDataset = useDeepCompareCallback(() => {
     if (searchParams.persistentId) {
-      return getDatasetByPersistentId(repository, searchParams.persistentId, searchParams.version)
+      return getDatasetByPersistentId(
+        repository,
+        searchParams.persistentId,
+        searchParams.version,
+        undefined,
+        true
+      )
     }
     if (searchParams.privateUrlToken) {
       return getDatasetByPrivateUrlToken(repository, searchParams.privateUrlToken)

src/sections/dataset/dataset-metadata/dataset-metadata-fields/DatasetMetadataFieldValueFormatted.tsx

@@ -1,3 +1,4 @@
+import TurndownService from 'turndown'
 import {
   METADATA_FIELD_DISPLAY_FORMAT_NAME_PLACEHOLDER,
   METADATA_FIELD_DISPLAY_FORMAT_PLACEHOLDER,
@@ -17,6 +18,13 @@ interface DatasetMetadataFieldValueFormattedProps {
   metadataFieldValue: DatasetMetadataFieldValueModel
   metadataBlockDisplayFormatInfo: MetadataBlockInfoDisplayFormat
 }
+
+const turndownService = new TurndownService()
+
+function transformHtmlToMarkdown(source: string): string {
+  return turndownService.turndown(source)
+}
+
 export function DatasetMetadataFieldValueFormatted({
   metadataBlockName,
   metadataFieldName,
@@ -36,6 +44,12 @@ export function DatasetMetadataFieldValueFormatted({
     t(`${metadataBlockName}.datasetField.${metadataFieldName}.name`)
   )
 
+  if (metadataBlockDisplayFormatInfo.fields[metadataFieldName]?.type === 'TEXTBOX') {
+    const markdownValue = transformHtmlToMarkdown(valueFormattedWithNamesTranslated)
+
+    return <MarkdownComponent markdown={markdownValue} />
+  }
+
   return <MarkdownComponent markdown={valueFormattedWithNamesTranslated} />
 }
 
@@ -80,9 +94,18 @@ function joinSubFields(
   metadataBlockInfo?: MetadataBlockInfoDisplayFormat
 ): string {
   return Object.entries(metadataSubField)
-    .map(([subFieldName, subFieldValue]) =>
-      formatSubFieldValue(subFieldValue, metadataBlockInfo?.fields[subFieldName]?.displayFormat)
-    )
+    .map(([subFieldName, subFieldValue]) => {
+      let formattedSubFieldValue = formatSubFieldValue(
+        subFieldValue,
+        metadataBlockInfo?.fields[subFieldName]?.displayFormat
+      )
+      const subFieldType = metadataBlockInfo?.fields[subFieldName]?.type as string | undefined
+
+      if (subFieldType === 'TEXTBOX')
+        formattedSubFieldValue = transformHtmlToMarkdown(formattedSubFieldValue)
+
+      return formattedSubFieldValue
+    })
     .join(' ')
 }
 

src/sections/shared/hierarchy/BreadcrumbsGenerator.tsx

@@ -45,7 +45,10 @@ export function BreadcrumbsGenerator({
             <Breadcrumb.Item
               key={index}
               linkAs={LinkToPage}
-              linkProps={{ page: Route.COLLECTIONS_BASE, children: <>{item.name}</> }}>
+              linkProps={{
+                page: Route.COLLECTIONS_BASE,
+                children: <>{item.name}</>
+              }}>
               {item.name}
             </Breadcrumb.Item>
           )

tests/component/metadata-block-info/domain/models/MetadataBlockInfoMother.ts

@@ -17,25 +17,25 @@ export class MetadataBlockInfoMother {
     return {
       name: MetadataBlockName.CITATION,
       fields: {
-        alternativePersistentId: { displayFormat: '' },
-        publicationDate: { displayFormat: '' },
-        citationDate: { displayFormat: '' },
-        title: { displayFormat: '' },
-        subject: { displayFormat: ';' },
-        author: { displayFormat: '' },
-        authorName: { displayFormat: '#VALUE' },
-        authorAffiliation: { displayFormat: '(#VALUE)' },
-        authorIdentifierScheme: { displayFormat: '- #VALUE:' },
-        authorIdentifier: { displayFormat: '[#VALUE](https://orcid.org/#VALUE)' },
-        datasetContact: { displayFormat: '#VALUE' },
-        datasetContactName: { displayFormat: '#VALUE' },
-        datasetContactAffiliation: { displayFormat: '(#VALUE)' },
-        datasetContactEmail: { displayFormat: '[#VALUE](mailto:#VALUE)' },
-        dsDescription: { displayFormat: '' },
-        dsDescriptionValue: { displayFormat: '#VALUE' },
-        producerURL: { displayFormat: '[#VALUE](#VALUE)' },
-        producerLogoURL: { displayFormat: '![#NAME](#VALUE)' },
-        dateOfCollectionStart: { displayFormat: '#NAME: #VALUE ' }
+        alternativePersistentId: { displayFormat: '', type: 'TEXT' },
+        publicationDate: { displayFormat: '', type: 'DATE' },
+        citationDate: { displayFormat: '', type: 'DATE' },
+        title: { displayFormat: '', type: 'TEXT' },
+        subject: { displayFormat: ';', type: 'TEXT' },
+        author: { displayFormat: '', type: 'NONE' },
+        authorName: { displayFormat: '#VALUE', type: 'TEXT' },
+        authorAffiliation: { displayFormat: '(#VALUE)', type: 'TEXT' },
+        authorIdentifierScheme: { displayFormat: '- #VALUE:', type: 'TEXT' },
+        authorIdentifier: { displayFormat: '[#VALUE](https://orcid.org/#VALUE)', type: 'TEXT' },
+        datasetContact: { displayFormat: '#VALUE', type: 'NONE' },
+        datasetContactName: { displayFormat: '#VALUE', type: 'TEXT' },
+        datasetContactAffiliation: { displayFormat: '(#VALUE)', type: 'TEXT' },
+        datasetContactEmail: { displayFormat: '[#VALUE](mailto:#VALUE)', type: 'EMAIL' },
+        dsDescription: { displayFormat: '', type: 'NONE' },
+        dsDescriptionValue: { displayFormat: '#VALUE', type: 'TEXTBOX' },
+        producerURL: { displayFormat: '[#VALUE](#VALUE)', type: 'NONE' },
+        producerLogoURL: { displayFormat: '![#NAME](#VALUE)', type: 'URL' },
+        dateOfCollectionStart: { displayFormat: '#NAME: #VALUE ', type: 'NONE' }
       },
       ...props
     }

tests/component/sections/dataset/dataset-metadata/DatasetMetadata.spec.tsx

@@ -323,5 +323,6 @@ describe('DatasetMetadata', () => {
     )
 
     cy.findAllByTestId('ds-metadata-block-display-format-error').should('exist')
+    cy.contains('Error getting metadata block display info').should('not.exist')
   })
 })